[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20140617155830.GE21752@arm.com>
Date: Tue, 17 Jun 2014 16:58:30 +0100
From: Catalin Marinas <catalin.marinas@....com>
To: Andrew Pinski <apinski@...ium.com>
Cc: linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 07/24] ARM64:ILP32: Use the same size and layout of the
signal structures for ILP32 as for LP64.
On Sat, May 24, 2014 at 12:02:02AM -0700, Andrew Pinski wrote:
> diff --git a/arch/arm64/include/uapi/asm/siginfo.h b/arch/arm64/include/uapi/asm/siginfo.h
> index 5a74a08..1a6aa32 100644
> --- a/arch/arm64/include/uapi/asm/siginfo.h
> +++ b/arch/arm64/include/uapi/asm/siginfo.h
> @@ -1,5 +1,6 @@
> /*
> * Copyright (C) 2012 ARM Ltd.
> + * Copyright (C) 2014 Cavium Inc.
> *
> * This program is free software; you can redistribute it and/or modify
> * it under the terms of the GNU General Public License version 2 as
> @@ -18,6 +19,26 @@
>
> #define __ARCH_SI_PREAMBLE_SIZE (4 * sizeof(int))
>
> +#ifdef __ILP32__
Does the compiler define __ILP32__ or it just undefined __LP64__.
Anyway, maybe we should have some consistency and only use one in all
places.
> +# ifdef __AARCH64EB__
> +# define __SIGINFO_INNER(type, field) \
> + int __pad#field; \
> + type field
> +# else
> +# define __SIGINFO_INNER(type, field) \
> + type field; \
> + int __pad#field
> +# endif
> +
> +# undef __SIGINFO_VOIDPOINTER
> +# define __SIGINFO_VOIDPOINTER(field) \
> + __SIGINFO_INNER(void __user*, field)
> +# undef __SIGINFO_BAND
> +
> +# define __SIGINFO_BAND(field) \
> + __SIGINFO_INNER(long, field)
> +#endif
> +
> #include <asm-generic/siginfo.h>
>
> #endif
Do the above give us the correct alignment for the _sifields union in
siginfo? Or we get it as a side-effect of some other structure member
forcing 64-bit alignment?
> diff --git a/arch/arm64/include/uapi/asm/signal.h b/arch/arm64/include/uapi/asm/signal.h
> index 8d1e723..d90d53b 100644
> --- a/arch/arm64/include/uapi/asm/signal.h
> +++ b/arch/arm64/include/uapi/asm/signal.h
> @@ -19,6 +19,38 @@
> /* Required for AArch32 compatibility. */
> #define SA_RESTORER 0x04000000
>
> +/* For ILP32, sigset should be the same size fields as LP64 so use
> + unsigned long long. */
> +#ifdef __ILP32__
> +#define __SIGSET_INNER_TYPE __extension__ unsigned long long
> +#define _NSIG_BPW 64
> +
> +# ifdef __AARCH64EB__
> +# define __SIGNAL_INNER(type, field) \
> + __extension__ struct { \
> + int __pad_##field; \
> + type field; \
> + } __attribute__((aligned(8)))
> +# else
> +# define __SIGNAL_INNER(type, field) \
> + __extension__ struct { \
> + type field; \
> + int __pad_##field; \
> + } __attribute__((aligned(8)))
> +# endif
> +
> +# define __SIGACTION_HANDLER(field) \
> + __SIGNAL_INNER(__sighandler_t, field)
> +
> +
> +#define __SIGACTION_FLAGS(field) \
> + __extension__ unsigned long long field
> +
> +#define __SIGACTION_RESTORER(field) \
> + __SIGNAL_INNER(__sigrestore_t, field)
> +
> +#endif
> +
> #include <asm-generic/signal.h>
>
> #endif
I haven't seen any UAPI header using __extension__. Is this allowed? I
assume we should be allowed to use the exported kernel headers with
non-gcc compilers.
--
Catalin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists